CODEWITHSUNDEEP
wxpythonmatplotlib
madtowneast
madtowneast

Reputation: 2390

Matplotlib - Grid arrangement subplots spacing

I am trying to arrange a bunch of subplots in a grid like fashion. The problem is that the number of subplots varies with user selection of what data to plot. Right now I am trying to add plots this way:

    l = len(clicked)

    self.p=wx.Panel(self)
    self.dpi=100
    self.fig = Figure()
    self.canvas = FigCanvas(self.p, -1, self.fig)
    if l == 1:
        self.splt = self.fig.add_subplot(1,1,1)
        self.plt=self.splt.plot(np.arange(5),np.arange(5))
    else:
        for i in np.arange(l):
            self.splt=self.fig.add_subplot(l+1,2,i+1)
            self.fig.subplots_adjust(left=0.7, bottom=0.6, right=0.75, top=0.75)
            self.plt=self.splt.plot(np.arange(5),np.arange(5))

I am just using fake data for debugging purposes. Anyhow, I am using wxPython to draw this inside a frame. clicked here provides the number of selections the user made. I already tried using subplots_adjust(), with quite the opposite result than I wanted. The plots are being shrunk something indiscernable. Is there a way to arrange teh plots in some sort of grid. I saw there is an option subplot2grid, but I havent gotten it to work with the variable number of subplots.

Upvotes: 1

Views: 754

Answers (1)

Daan
Daan

Reputation: 2059

Hopefully this helps:

from __future__ import division
import numpy as np
import pylab as plt

def divideSquare(N):
    if N<1:
        return 1,1

    minX = max(1,int(np.floor(np.sqrt(N))))
    maxX = max(minX+1, int(np.ceil(np.sqrt(5.0/3*N))))
    Nx = range(minX, maxX+1)
    Ny = [np.ceil(N/y) for y in Nx]
    err = [np.uint8(y*x - N) for y in Nx for x in Ny]
    ind = np.argmin(err)
    y = Nx[int(ind/len(Ny))]
    x = Ny[ind%len(Nx)]
    return min(y,x), max(y,x)

Nlist = np.arange(0,101)
empty = np.zeros_like(Nlist)
ratio = np.zeros_like(Nlist, dtype = float)
for k, N in enumerate(Nlist):
    y,x = divideSquare(N)
    empty[k] = y*x - N
    ratio[k] = 1.0 * x/y
    print y,x,y/x

plt.figure(1)
plt.clf()
y,x = divideSquare(2)
plt.subplot(y,x,1)    
plt.plot(Nlist, empty, 'k')
plt.xlabel('Number of plots')
plt.ylabel('Empty squares left')
plt.subplot(y,x,2)
plt.plot(Nlist, ratio, 'r')
plt.xlabel('Number of plots')
plt.ylabel('Ratio')

def divide2or3(N):
    if N<1:
        return 1,1
    if N%2==0:
        return int(np.ceil(N/2)), 2
    return int(np.ceil(N/3)), 3

Nlist = np.arange(0,101)
empty = np.zeros_like(Nlist)
ratio = np.zeros_like(Nlist, dtype = float)
for k, N in enumerate(Nlist):
    y,x = divide2or3(N)
    empty[k] = y*x - N
    ratio[k] = 1.0 * x/y
    print y,x,y/x

plt.figure(2)
plt.clf()
y,x = divide2or3(2)
plt.subplot(y,x,1)    
plt.plot(Nlist, empty, 'k')
plt.xlabel('Number of plots')
plt.ylabel('Empty squares left')
plt.subplot(y,x,2)
plt.plot(Nlist, ratio, 'r')
plt.xlabel('Number of plots')
plt.ylabel('Ratio')

plt.show()

divideSquare(N) attempts to get a ratio close to 1 but with a minimal number of empty positions. divide2or3(N) gives you a grid of 2 or 3 columns (or rows if you invert x and y), but I'm not sure if this is what you want for 90 plots.

Upvotes: 2

Related Questions